Matches in SemOpenAlex for { <https://semopenalex.org/work/W4386977312> ?p ?o ?g. }
Showing items 1 to 59 of
59
with 100 items per page.
- W4386977312 abstract "We study offline model-based optimization to maximize a black-box objective function with a static dataset of designs and scores. These designs encompass a variety of domains, including materials, robots and DNA sequences. A common approach trains a proxy on the static dataset to approximate the black-box objective function and performs gradient ascent to obtain new designs. However, this often results in poor designs due to the proxy inaccuracies for out-of-distribution designs. Recent studies indicate that: (a) gradient ascent with a mean ensemble of proxies generally outperforms simple gradient ascent, and (b) a trained proxy provides weak ranking supervision signals for design selection. Motivated by (a) and (b), we propose textit{parallel-mentoring} as an effective and novel method that facilitates mentoring among parallel proxies, creating a more robust ensemble to mitigate the out-of-distribution issue. We focus on the three-proxy case and our method consists of two modules. The first module, textit{voting-based pairwise supervision}, operates on three parallel proxies and captures their ranking supervision signals as pairwise comparison labels. These labels are combined through majority voting to generate consensus labels, which incorporate ranking supervision signals from all proxies and enable mutual mentoring. However, label noise arises due to possible incorrect consensus. To alleviate this, we introduce an textit{adaptive soft-labeling} module with soft-labels initialized as consensus labels. Based on bi-level optimization, this module fine-tunes proxies in the inner level and learns more accurate labels in the outer level to adaptively mentor proxies, resulting in a more robust ensemble. Experiments validate the effectiveness of our method. Our code is available here." @default.
- W4386977312 created "2023-09-23" @default.
- W4386977312 creator A5010239302 @default.
- W4386977312 creator A5031092503 @default.
- W4386977312 creator A5033684035 @default.
- W4386977312 creator A5048279362 @default.
- W4386977312 creator A5075885606 @default.
- W4386977312 date "2023-09-20" @default.
- W4386977312 modified "2023-10-12" @default.
- W4386977312 title "Parallel-mentoring for Offline Model-based Optimization" @default.
- W4386977312 doi "https://doi.org/10.48550/arxiv.2309.11592" @default.
- W4386977312 hasPublicationYear "2023" @default.
- W4386977312 type Work @default.
- W4386977312 citedByCount "0" @default.
- W4386977312 crossrefType "posted-content" @default.
- W4386977312 hasAuthorship W4386977312A5010239302 @default.
- W4386977312 hasAuthorship W4386977312A5031092503 @default.
- W4386977312 hasAuthorship W4386977312A5033684035 @default.
- W4386977312 hasAuthorship W4386977312A5048279362 @default.
- W4386977312 hasAuthorship W4386977312A5075885606 @default.
- W4386977312 hasBestOaLocation W43869773121 @default.
- W4386977312 hasConcept C119857082 @default.
- W4386977312 hasConcept C153668964 @default.
- W4386977312 hasConcept C154945302 @default.
- W4386977312 hasConcept C17744445 @default.
- W4386977312 hasConcept C184898388 @default.
- W4386977312 hasConcept C189430467 @default.
- W4386977312 hasConcept C199539241 @default.
- W4386977312 hasConcept C2780148112 @default.
- W4386977312 hasConcept C41008148 @default.
- W4386977312 hasConcept C520049643 @default.
- W4386977312 hasConcept C94625758 @default.
- W4386977312 hasConceptScore W4386977312C119857082 @default.
- W4386977312 hasConceptScore W4386977312C153668964 @default.
- W4386977312 hasConceptScore W4386977312C154945302 @default.
- W4386977312 hasConceptScore W4386977312C17744445 @default.
- W4386977312 hasConceptScore W4386977312C184898388 @default.
- W4386977312 hasConceptScore W4386977312C189430467 @default.
- W4386977312 hasConceptScore W4386977312C199539241 @default.
- W4386977312 hasConceptScore W4386977312C2780148112 @default.
- W4386977312 hasConceptScore W4386977312C41008148 @default.
- W4386977312 hasConceptScore W4386977312C520049643 @default.
- W4386977312 hasConceptScore W4386977312C94625758 @default.
- W4386977312 hasLocation W43869773121 @default.
- W4386977312 hasOpenAccess W4386977312 @default.
- W4386977312 hasPrimaryLocation W43869773121 @default.
- W4386977312 hasRelatedWork W1894159578 @default.
- W4386977312 hasRelatedWork W2011472225 @default.
- W4386977312 hasRelatedWork W2055572829 @default.
- W4386977312 hasRelatedWork W2767338541 @default.
- W4386977312 hasRelatedWork W2901018557 @default.
- W4386977312 hasRelatedWork W3000057026 @default.
- W4386977312 hasRelatedWork W3048565508 @default.
- W4386977312 hasRelatedWork W3121841074 @default.
- W4386977312 hasRelatedWork W3163984363 @default.
- W4386977312 hasRelatedWork W4313908216 @default.
- W4386977312 isParatext "false" @default.
- W4386977312 isRetracted "false" @default.
- W4386977312 workType "article" @default.